https://bugs.kde.org/show_bug.cgi?id=495817
--- Comment #13 from Florian Krohm <flo2...@eich-krohm.de> --- In bbebdbb2b1094b08ccbd8d09fc65f6fa0b9fecc0 I checked in the "final bits to fix disassembly for vector insns" patch. With that "disasm-test --all --summary --show-spec-exc" reports: *** mismatch VEX: |nop 0(%r4,%r0)| objdump: |nop 0(%r4| *** mismatch VEX: |nop 1(%r4,%r0)| objdump: |nop 1(%r4| *** mismatch VEX: |nop 2(%r12,%r0)| objdump: |nop 2(%r12| *** mismatch VEX: |nop 4095(%r2,%r0)| objdump: |nop 4095(%r2| *** specification exception for insn E66000000007 in vler.dump *** specification exception for insn E66040000007 in vler.dump .... *** specification exception for insn E7640002304D in vrep.dump *** specification exception for insn E764FFFF004D in vrep.dump Total: 16163 tests generated Total: 16142 insns verified Total: 4 disassembly mismatches Total: 21 specification exceptions This disasm-test was using objdump 2.38 under the covers. The mismatches are due to a bug in binutils which has been fixed in the meantime. The specification exceptions occur because not all opcode constraints can be expressed in the test generator. With objdump from current git af829d8b8d5bb0ce658bbad6d33f69dcb534343b there are many miscompares most of which are of this kind: *** mismatch VEX: |bih 1(%r11,%r0)| objdump: |bih 1(%r11,0)| i.e. GPR #0 as base register is now disassembled as 0, formerly %r0. Other miscompares are for the R*SBG family of opcodes. -- You are receiving this mail because: You are watching all bug changes.
